Current state-of-the-art femtosecond solid-state lasers, such as titanium-sapphire amplification systems, have no problem generating high peak powers. The benchmark laser system is BELLA, producing 30 fs pulses with 40 J of pulse energy, i.e. >1 PW peak power, at a repetition rate of 1 Hz, resulting in an average power of 40 W. However, such laser systems are far from compact, with sizes approaching that of a building rather than a table, or efficient with wall-plug efficiencies in the range of 10-4. Moreover, a scaling of the average power and thus the repetition rate is inhibited by thermo-optical problems.
Only a higher average power and repetition rate, however, can bring these laboratory- tested concepts to a facility in which a broad range of users and their industrial, scientific and medical applications will benefit. There are two general approaches to extend the achievable parameter field of current laser sources toward this goal.

Peak power evolution of ultrafast fiber laser systems
Peak power evolution of ultrafast fiber laser systems; blue: single fiber systems; blue line: theoretical performance limit of single emitter; red: systems based on coherent combination.

On the one hand, a great deal of work is being devoted to increase the performance parameters of single fiber systems. This approach includes increasing the repetition rate and therefore the average power of state-of-the-art bulk-laser technology already emitting extreme peak powers as well as improving the peak-power capability of laser architectures that have already been proven to handle high average powers like slabs, thin-disks and fibers. Ytterbium-based femtosecond fiber amplifiers represent a particularly promising approach due to their simple and efficient single-pass setups, the broad spectral bandwidth of ytterbium-silica supporting sub-200 fs pulse duration and their capability to deliver high average powers with an excellent beam quality. Novel rare-earth-doped fiber designs have enabled significant mode-area scaling. Worthy of mention, ytterbium-doped photonic-crystal fibers, especially in combination with a rod-type architecture, lifted the record peak power of high repetition rate ultrafast fiber lasers from a few MW to the GW level in 2005 to 2007.
Scaling concept independent of the amplifier technology has been recently demonstrated using parallelization of multiple ultrafast amplifiers and a subsequent coherent combination of the beams. In this approach, the output of a laser front-end is divided into multiple spatially separated channels. In each channel, the amplifier can now be pushed to its specific limitations. Such a setup thus essentially represents an amplifying interferometer. In the case of perfect recombination, the total pulse energy and average power results from the sum of each of the individual channels. Due to their simple single-pass setup, reproducible beam profile and amplification characteristics, fiber lasers are ideal candidates for this spatial multiplexing concept. So far, this approach has been demonstrated for up to four femtosecond fiber amplifiers in the CPA regime with peak-powers of up to 22 GW (5.7 mJ at 40 kHz, 200 fs).

Border Bottom